Statistics Seminar - 10/10/23

Abstract:

Label-free quantification is a fast-growing methodology to infer abundances in mass spectrometry proteomics. Extensive research has focused on spectral quantification and peptide identification. However, research towards modeling and understanding quantitative proteomics data is scarce. We propose here a Bayesian hierarchical decision model (Baldur) to test for differences in means between conditions for proteins, peptides, and post-translation modifications. We develop a Bayesian regression model to characterize local mean-variance trends in the data, to describe measurement uncertainty, and to estimate the decision model hyperparameters. A key contribution is the development of a new gamma regression model that describes the mean-variance dependency as a mixture of a common and a latent trend—allowing for localized trend estimates. We then evaluate the performance of Baldur, limma-trend and t-test on six enchmark datasets: five total proteomics and one post-translational modification dataset. We find that Baldur drastically improves the decision in noisier post-translational modification data over limma-trend and t-test. In addition, we see significant improvements using Baldur over the other methods in the total proteomics datasets. Finally, we analyzed Baldur’s performance when increasing the number of replicates and found that the method always increases precision with sample size while showing robust control of the false positive rate. We conclude that our model vastly improves over popular data analysis methods (limma-trend and t-test) in several spike-in datasets by achieving a high true positive detection rate while greatly reducing the false positive rate.

Bio:

Philip Berg is a Postdoctoral Researcher in computational biology at IGBB, Mississippi State University. He has obtained a PhD in Biochemistry and Life Sciences from The Mississippi State University while following a course work in Statistics and Bioinformatics. He holds a MSc in Genomics and Systems Biology from the University of Gothenburg and a BSc in Plant Biology from the Swedish University of Agriculture.

While at IGBB, Philip has contributed to two NSF research projects in computational and systems biology, generating research results for several papers published in BMC Bioinformatics and Plant Physiology. He has presented his research at several conferences including RECOMB, CSHL Probabilistic Modeling in Genomics, MCBIOS, EMBO Systems Biology Workshop and Plant Biology 2023. He is the author of baldur, a CRAN package for Bayesian data analysis of label-free proteomics data.